Insulin resistance (IR) is defined as decreased tissue response to insulin-mediated cellular actions and is the inverse of insulin sensitivity.1 In adults, IR has been associated with obesity, metabolic syndrome, hypertension, and cardiovascular disease.2 In addition, it is recognized as a determining condition for the development of type 2 diabetes mellitus (T2D). The scenario is similar among children, since studies have shown that IR is significantly associated with obesity and cardiometabolic risk.3

The assessment of IR is clinically relevant and can be performed by several methods. Hyperinsulinemic-euglycemic clamp is the gold standard test to diagnose IR. However, it is expansive, invasive, and not easily available in clinical settings.4 An alternative and widely used method in clinical practice is the homeostasis model assessment for insulin resistance (HOMA-IR). However, HOMA-IR also depends on fasting blood collection and plasma insulin assay, which is expensive and not available in many clinics dedicated to primary care.5 Hence, cheaper biomarkers such as the triglycerides/glucose index (TyG index)6 and the triglycerides-to-HDL-C ratio (TG/HDL-C)7 have been proposed.

Cutoff values for HOMA-IR in pediatric populations are quite variable.8 According to the results of a systematic review, of all studies that established cutoffs using ROC curve, the lowest cutoffs for HOMA-IR reported in the literature were 1.65 and 1.95 for girls and boys, respectively, whereas the highest were 3.82 and 5.22.9 Moreover, a Korean study conducted with data from 3313 children and adolescents, aged 10–18 years, established optimum cutoff values for HOMA-IR and the TyG index to provide screening for insulin resistance according to three different criteria.10 Dependent on the criterion for metabolic syndrome, cutoffs for HOMA-IR varied from 2.96 to 3.69 for girls and from 2.86 to 3.54 for boys.

Although obtaining lipid fractions and glucose levels is relatively cheaper than obtaining insulin levels to calculate HOMA-IR, biochemical exams always require some cost, besides the need for fasting blood collection.

Studies on the effectiveness of biochemical markers compared to anthropometric indicators to provide screening for IR are scarce in the literature. Therefore, given the importance of incorporating simple and low cost tools into the pediatric clinical setting to provide screening for IR, the present study sought to investigate whether waist-to-height ratio (WHtR) is comparable to biochemical indicators for the discrimination of IR in Brazilian children and adolescents.

Clinical, demographic, anthropometric, and biochemical data stratified by sex are summarized in Table 1. TG, TG/HDL-C ratio, TyG index, insulin, HOMA-IR, and IR percentage were higher in girls, while SBP was higher in boys. In addition, the proportion of IR observed was 32.9% among girls and 18.5% among boys (p<0.001).

Table 2 shows the clinical, demographic, anthropometric, and biochemical data stratified by IR status for both sexes. The proportion of overweight, SBP, TG, TG/HDL-C ratio, TyG index, fasting insulin, glucose, HOMA-IR, and uric acid were higher in the insulin resistant of both sexes.

Partial correlation coefficients are summarized in Table 3. In girls, HOMA-IR was more strongly correlated with WHtR than with TyG index and TG/HDL-C (r=0.60 vs. r=0.42 and r=0.39, respectively). In boys, HOMA-IR was weakly correlated with TG/HDL-C and moderately correlated with both WHtR and TyG index (r=0.56 and r=0.50, respectively).

Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and AUC are exhibited for both sexes in Table 4. In girls, the highest true positive rate (sensitivity) was obtained by TG/HDL-C. In turn, the true negative rate (specificity) of TG/HDL-C was the lowest among the indices. Differently in boys, WHtR and TyG index reached more balanced sensitivity and specificity compared to TG/HDL-C. For both boys and girls, the discriminatory power for IR was similar between the TyG index, TG/HDL-C ratio, and WHtR.

The main finding of this study was that the TyG index, TG/HDL-C, and WHtR provided similar discriminatory power for IR. In addition, the data show that overweight girls and boys have a greater probability of being insulin resistant and dyslipidemic than those with normal weight.

Studies have shown that IR is a risk factor for the development of T2D and cardiovascular disease in children and adolescents.1,3 In the present study, the frequency of overweight was greater in insulin resistant than in non-insulin resistant girls and boys. In addition, those with IR presented with higher values of BMI, WC, WHtR, TG, TG/HDL-C, TyG index, fasting insulin and glucose, serum uric acid, and HOMA-IR. Corroborating these findings, Kang et al.,20 studying 221 Korean adolescents, reported higher values of BMI, WC, TG, TG/HDL-C, TyG index, fasting insulin and glucose, and HOMA-IR, besides lower HDL-C levels, in those diagnosed with IR. Similarly, Iwani et al.21 reported higher values of TG, LDL-C, TG/HDL-C, BMI, and WC in overweight and obese insulin resistant children and adolescents.

In the last years, the WHO has recommended the use of anthropometric indicators to identify individuals with cardiometabolic risk. Looking in this direction, Behboudi-Gandevani et al., 2016 have recently shown that WHtR was a good predictor of insulin resistance and metabolic syndrome in women with polycystic ovary syndrome, which could advocate WHtR as a promising screening tool for this disease.22 Specifically in a pediatric population, Ibarra-Reynoso et al.23 studied the factors associated with whole body and hepatic IR as estimated by insulin-like growth factor binding protein-1 (IGFBP-1) in children aged 6–11 years. In that study, obesity was classified according to BMI, visceral adiposity, and WHtR. Interestingly, for those obese children, the predictors of HOMA-IR (whole body IR) were the same independently of the indicator used to provide obesity diagnosis. Moreover, a reduction of 2.15ng/mL in the values of IGFBP-1 was estimated for each 1cm of increase in WHtR, thus indicating that WHtR was a strong predictor of hepatic IR.

To the best of the authors’ knowledge, the present study was the first to compare the effectiveness of WHtR to provide screening for IR in a pediatric population. The results showed that all provided moderate correlation with HOMA-IR in both sexes (Table 3). Corroborating these findings, Guerrero-Romero et al.,7 studying the general population of Guadalajara, reported moderate correlation between the TyG index and HOMA-IR (r=0.39). In the study by Kang et al.,20 moderate correlation was also observed for both the TyG index and TG/HDL-C with HOMA-IR (r=0.41 and r=0.40, respectively). These results indicate that both the TyG index and TG/HDL-C are equally associated with HOMA-IR in pediatric populations of different ages. In the present study, WHtR had a stronger correlation with HOMA-IR than the other indicators, mainly among girls.

Regarding the analysis of the ROC curve, no indicator stood out from the others in the comparison of AUC, thus indicating similar discriminatory power for IR between WHtR and indices based on biochemical markers, even adjusting for age. Partially corroborating these findings, Kang et al.20 observed similar discriminatory power for IR when comparing the TyG index and TG/HDL-C. However, the present study was the first to show that the WHtR was similar to the TyG index and TG/HDL-C to discriminate IR in a pediatric population.

Consistent evidence has documented the role of abdominal fat deposits in the pathogenesis of metabolic disorders, including insulin resistance.24 In this sense, the biochemical predictors of IR, which are based on altered blood lipids, are probably collinear to abdominal adiposity. Corroborating this assumption, Silva et al.25 have shown that WHtR was more strongly correlated (r=0.78) with trunk fat as measured by DEXA than other anthropometric indicators, including waist circumference and waist-to-hip ratio. Moreover, Mazidi et al.26 recently approached the complex link between glucose/insulin homeostasis and serum uric acid. Using a regression analysis with a structural equation model, the authors found that adiposity factors such as visceral adipose tissue, BMI, and WC are mediators of the association between insulin resistance and serum uric acid. In the present study, the girls and boys in the highest quartile for WHtR presented with higher HOMA-IR compared to those in the lowest quartiles, as occurred with the TyG index and TGC/HDL-C as well (Supplementary Table 1). Indeed, this indirect evidence supports WHtR as a proxy of regional body fat, at least in the studied population. This information is very relevant to the public health system because WHtR is easy to determine in a clinical setting. Besides, unlike the biochemical markers, WHtR can be used in absence of invasive fasting laboratory exams, gives instant results, is free of costs, and is relatively easy to use.

This study has some limitations. First, all volunteers belong to nine public schools of the municipality of Vitória, so these results cannot be extrapolated to the general population of children and adolescents living in Brazil. Second, HOMA-IR is a validated method widely used in clinical practice for the diagnosis of IR; however, it would be helpful if we could evaluate the discriminatory power of the indicators, using hyperinsulinemic-euglycemic clamp as a reference (gold standard test). An appropriate sample size, which allowed for high statistical power, and the fact that anthropometric measures were obtained by a single investigator, were the main strengths of this study.

In conclusion, WHtR reached similar discriminatory power to biochemical markers for IR in children and adolescents. Thus, taking into account its cost–benefit ratio, it is plausible to suggest that WHtR may be a useful tool to provide screening for IR in pediatric populations.

Pro-Rectory of Research and Postgraduation of the Federal University of Espírito Santo (Institutional Program Fund for Research Support – 2015); FAPES/PPSUS (grant number 65854420/2014); Fundação VALE; Laboratory Thommasi of Vitória, which funded the accomplishment of part of the biochemical exams. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The authors declare no conflicts of interest.